噪声背景下检测突变信息的奇异值分解技术

研究了利用信号时间序列重构的吸引子轨迹矩阵奇异值分布特征与信号自身特征的关系,提出了基于选择轨迹矩阵中主要反映突变信息明显的奇异值进行信号重构的方法来检测信号中的突变信息,并对重构奇异值的选择进行了研究,给出了简便的方法.该方法旨在同时降低光滑信号和噪声的影响,从而更加清晰地表征突变信息.利用该方法对仿真信号的处理结果,并将检测结果与未改进之前的检测结果进行了比较,结果表明,改进后的奇异值分解技术比改进前有更大的信噪比,能够更有效地应用于信号的奇异性检测.并成功地将该方法用于旋转机械静动件较轻碰摩和较重碰摩故障实验信号的检测,获得了满意的结果.

An improved method of detecting abrupt information based on singularity value decomposition in noise background

Based on the relations between the time series and the singular value distribution of the Singularity Value Decomposition about track matrix of attractor reconstructed by time series, a method of detecting abrupt information in the signal is presented by selecting most potential reflecting singular values for reconstruction of signal. The selective method of singular values for reconstruction is also gained. The abrupt information is easier to be sized because satin signal and noise can be reduced at the same time by using the method. The method is performed according to several numerical signals. Through comparing its detection results with ones gotten using the unimproved method, the conclusion can be obtained that the improved method is feasible and higher signal-to-noise ratio, so it is good at detecting signal singularity. Finally, the method is successful to be used to detect the test signals of gentle rub-impact and heavy rub-impact between rotor and stator of rotating machinery, and the satisfactory results are gotten.